Modeling the potential impact of abuse-deterrent opioids on medical resource utilization

Abstract

Objectives: To extend a previously published manuscript on a model for estimating potential avoided medical events and cost savings in the US associated with the introduction of extended-release abuse-deterrent opioids and incorporate new methods of evaluating abuse deterrence using human abuse potential studies.

Methods: A model was developed to estimate reductions in abuse-related events and annual savings in the US. Model inputs included: opioid abuse prevalence, abuse-deterrent opioid cost and effectiveness at deterring abuse, and opioid abuse-related events and costs. Direct (medical and drug) and indirect (work loss) cost savings (2017 US$) and abuse-related events were estimated assuming the replacement of the entire extended-release opioid market (brand and generic) by brand abuse-deterrent opioids.

Results: Replacing the extended-release opioid market with abuse-deterrent opioids is estimated to lower annual abuse-related medical events by ∼13–31% (e.g. 78,000–186,000 emergency department visits) and lower annual medical costs by ∼$640 M–$1,538 M, depending on the abuse-deterrent technology (physical/chemical barrier or agonist/antagonist). Replacement of extended-release oxycodone with extended-release abuse-deterrent oxycodone is associated with the largest amount of cost savings and highest number of avoided medical events, followed by replacing extended-release morphine with an extended-release abuse-deterrent opioid. Replacement of transdermal fentanyl is associated with the smallest amount of cost savings and lowest number of avoided medical events.

Conclusion: Agonist/antagonist abuse-deterrent opioid technology is associated with higher annual medical cost savings and more avoided events than physical/chemical barrier technology. Total net savings are dependent upon the abuse-deterrent opioid price relative to non-abuse-deterrent opioids.

Keywords:

• Opioid
• abuse
• abuse-deterrent formulation
• savings

JEL CLASSIFICATION CODES:

• I00
• I10

Introduction

Prescription opioid analgesic products (pain relievers) are widely regarded as the standard of care for the management of acute pain and chronic pain due to advanced medical illness¹. At the same time, prescription opioid abuse has become a national epidemic². In the US, drug overdose is the leading cause of accidental death, with opioids responsible for more than six out of 10 drug overdose fatalities³. The total economic burden of opioid abuse, estimated by the Centers for Disease Control and Prevention in 2016, is $78.5 billion per year⁴.

Oral ingestion (swallowing with or without chewing) remains the most common prescription opioid route of abuse, but some abusers also tamper (manipulate) with prescription opioids to allow for non-oral administration (e.g. inhalation, injection)⁵. A number of studies have demonstrated that administering oral opioids through non-oral routes is associated with significantly more severe adverse health consequences as compared to oral routes of abuse^5,6. One strategy designed to mitigate opioid abuse is the introduction of abuse-deterrent opioids. To date, the technologies address abuse that involves manipulation (i.e. crushing, chewing, or dissolving), irrespective of the route of abuse. In the last few years, several extended-release opioids have received “abuse-deterrent” labeling from the FDA⁷. These abuse-deterrent opioids have been shown to demonstrate a reduction in real-world abuse and total costs to the healthcare system⁸. There are two main classes of abuse-deterrent technologies that manufacturers use to deter abuse of their medications; physical/chemical barrier (PC) and agonist/antagonist. PC technology physically alters opioid pills/capsules to make them less conducive to physical and/or chemical manipulation and liquid extraction. These technologies make the drug more difficult to crush, and/or chew, and/or dissolve, and aim to deter abusers from manipulating the product in order to abuse via oral and/or non-oral routes (i.e. chewing, snorting, injecting). Agonist/antagonist technology involves adding an opioid receptor antagonist to the opioid product that aims to counteract the euphoric effect that the abuser seeks when the opioid is manipulated and taken in a manner other than as prescribed. Real-world studies have shown that both kinds of formulations are associated with reductions in real-world abuse as well as fewer adverse events and lower healthcare costs^8–10.

Previous work by White et al.¹¹ has estimated potential cost savings associated with a hypothetical abuse-deterrent opioid from a private-payer perspective. White et al.¹² have also quantified the association between positive subjective measures from human abuse potential studies and non-medical use of abuse-deterrent opioids¹². Positive subjective measures quantify the experience of drug liking and drug high from various drugs among study participants. The positive subjective measures score includes both positive and negative experiences with the drug. Reductions in the positive subjective measure of an abuse-deterrent opioid from the non-abuse-deterrent opioid have been correlated with reductions in non-medical use¹².

The opioid market has changed significantly since the development of the abuse-deterrent opioids cost savings manuscript published by White et al.¹¹ with greater prevalence of misuse and/or abuse, the release of several abuse-deterrent opioids^11,13, and greater scrutiny of opioid prescribing¹⁴. During the 12-month period between July 2016 and June 2017, three molecules comprised 87% of all extended-release opioid prescribing: oxycodone, morphine, and transdermal fentanyl¹⁵. However, only 24% of all extended-release prescriptions were for molecules with FDA-approved abuse-deterrent labeling. Previous work by Rossiter et al.⁹ found that the introduction of abuse-deterrent oxycodone was associated with annual savings of $430 M in healthcare costs. The current study expands on the budget impact model presented by White et al.¹¹ and analyzes the effects of completely replacing non-abuse-deterrent opioids with abuse-deterrent opioids for certain molecules. Specifically, the objective of the current study was to assess the impact on non-medical use rates and medical events and costs by replacing extended-release oxycodone, extended-release morphine, and extended-release transdermal fentanyl with extended-release abuse-deterrent opioids. The findings of the current study help quantify the magnitude and drivers of potential avoided misuse and/or abuse-related medical events and associated cost savings by prescribing abuse-deterrent opioids instead of non-abuse-deterrent opioids.

Methods

Overview

The previously published model by White et al.¹¹ was enhanced to (1) reflect recent changes in the opioid market, (2) estimate the reduction in misuse and/or abuse events due to extended-release abuse-deterrent opioids with available positive subjective measures scores from human abuse potential studies, and (3) account for the potential of individuals to divert to illegally obtained prescription opioids or illicit drugs (e.g. heroin). Misuse and/or abuse events and costs of the effect of extended-release abuse-deterrent opioids were modeled for the entire US population using two abuse-deterrent technologies:

• Physical/chemical barrier (PC) technology (i.e. abuse-deterrent opioid oxycodone); and

• Agonist/antagonist (AA) technology (i.e. abuse-deterrent opioid morphine sulfate/naltrexone hydrochloride).

Morphine, oxycodone, and transdermal fentanyl were used to model the extended-release opioid market as they comprised the great majority of extended-release opioids that could be replaced by abuse-deterrent opioids (methadone is a naturally long-acting opioid and could not be replaced with abuse-deterrent opioids technology). Extended-release oxymorphone, extended-release hydrocodone, and extended-release hydromorphone together made up less than 13% of the extended-release opioid market between July 2016 and June 2017 and were not included¹⁵. The number of avoided misuse and/or abuse-related events and associated costs of replacing all non-abuse-deterrent opioids extended-release morphine and transdermal fentanyl products with an extended-release abuse-deterrent opioid was assessed. Consistent with the White et al.¹¹ study, potential cost savings and avoided medical events due to misuse and/or abuse were calculated based on the excess annual healthcare utilization among a commercially-insured population as well as the proportion and costs of avoided abuse-related episodes. Potential cost savings were calculated as the product of the number of avoided events and the cost per event. The number of avoided misuse and/or abuse-related events and associated costs of replacing all non-abuse-deterrent opioids oxycodone products with an extended-release abuse-deterrent opioid was assessed using findings from the Rossiter et al.⁹ study. The total number of opioid-related emergency department visits was taken from a 2009 publication by the Drug Abuse Warning Network (DAWN), which reported that, prior to abuse-deterrent opioid introduction, ∼600,000 emergency department visits annually involved opiate/opioid analgesics¹⁶. The number of annual avoided emergency department visits was estimated by multiplying the reduction in non-medical use by the overall number of opioid-related emergency department visits.

Model inputs

In addition to the data sources outlined in White et al.¹¹ the inputs included opioid abuse prevalence, abuse-deterrent opioid effectiveness, and costs associated with opioid abuse-related events, and were based on previously published administrative claims database analyses, substance abuse treatment center data, and national survey data¹¹. The excess number of misuse and/or abuse-related events per diagnosed opioid abuser per year (e.g. excess number of emergency department visits, excess hospitalization) and healthcare costs per misuse and/or abuse-related event for a diagnosed opioid abuser (e.g. average cost per emergency department visits, average cost per hospitalization) were calculated using the White et al.¹¹ model and adjusted for inflation (costs were inflated to 2017 USD)¹¹. Drug-specific data on non-medical use of extended-release prescription opioids and their routes of administration were used to determine the prevalence of misuse and/or abuse that could be potentially deterred by an extended-release abuse-deterrent opioid (e.g. injection, snorting, chewing)¹¹. Data from clinical studies published on FDA-approved drug labels were used to estimate an extended-release abuse-deterrent opioid’s effectiveness in reducing non-medical use^17,18. Additionally, the following data sources were updated for the current study:

• Population statistics: 2016 US Census data, Medicare and Medicaid enrollment data were used to update the US population¹⁹.

• Non-medical use data: Data on non-medical use was taken from the 2014 National Survey on Drug Use and Health (NSDUH)²⁰.

• ER opioid market: Prescription volumes and costs were determined using data from the IMS National Prescription Audit and National Sales Perspective datasets from Q3 2016 to Q2 2017¹⁵.

Reduction in the rate of non-medical use for abuse-deterrent opioids

Figure 1 summarizes the approach to estimate reductions in rates of non-medical use associated with abuse-deterrent opioids. Differences in positive subjective measures, measured by the mean maximum effect (E_max) between new entrant abuse-deterrent opioids and comparator non-abuse-deterrent opioids was used to estimate the degree to which the abuse-deterrent opioids may resist intentional or unintentional misuse and/or abuse. The positive subjective measures score for drug liking includes both positive and negative experiences on a bipolar visual analog scale. Regressions from a White et al.¹² study showed that a 5-point reduction in overall drug liking was associated with a 0.10% decrease in the NSDUH lifetime non-medical use rates¹². Positive subjective measures scores from FDA approved drug labels for oxycodone (PC) and morphine/naltrexone (AA) were used to estimate the reduction in non-medical use due to switching from a non-abuse-deterrent opioid to an abuse-deterrent opioid. The non-medical use reduction associated with agonist/antagonist technology was 30.6% and with PC technology was 12.7%.

Figure 1. Estimated reductions in non-medical use. Abbreviations. ADO, abuse-deterrent opioid; E_max, mean maximum effect; NMU, non-medical use; NSDUH, National Survey on Drug Use and Health. This figure reflects calculation steps for agonist/antagonist ADOs only. Physical barrier ADOs have a drug liking E_max of 80.4 mm, resulting in a 12.7% reduction in predicted NMU rate. The difference in drug liking E_max was converted to point reductions in NMU according to White et al.¹² and used to calculate the percent reduction in predicted NMU rate. Notes: (1) ‘mm’ represents millimeters, which are one unit on the Drug Liking scale measured from 0–100. (2) The 30.6 percent reduction in the predicted NMU rate associated with replacement of a non-ADO with an ADO is calculated by dividing the predicted percentage point reduction in the NMU rate associated with replacement of a non-ADO with an ADO (∼0.43 percentage points) by the NMU rate associated with non-ADO as reported by the NSDUH (∼1.40 percentage points).

The prescribing of abuse-deterrent opioids may cause some patients to divert to non-abuse-deterrent opioids as well as illicit drugs. Two methods of diversion were considered in this current model: (1) non-abuse-deterrent opioids through illegal channels and (2) obtaining illicit drugs through illegal channels. For obtaining non-abuse-deterrent opioids through illegal channels, Rice et al.²¹ determined that 10% of prescription opioid abusers were obtaining their prescriptions through illegal channels and did not have evidence of any opioid prescriptions by them or immediate family members. For diversion of illegal drugs, Yenikomshian et al.¹⁰ estimate that 5% of prescription opioid abusers will divert to abusing illicit drugs (e.g. heroin)¹⁰.

Medical cost savings for extended-release oxycodone were taken from a Rossiter et al.⁹ publication estimating the potential cost savings associated with the replacement of non-abuse-deterrent oxycodone with PC technology abuse-deterrent oxycodone. To calculate the potential cost savings related to abuse-deterrent oxycodone with agonist/antagonist technology, the medical cost savings from Rossiter et al.⁹ were scaled up by a factor of 2.4, the ratio between the reduction in non-medical use for agonist/antagonist abuse-deterrent opioids (30.6%) and PC abuse-deterrent opioids (12.7%). The overall number of opioid related emergency department visits was multiplied by the rate of reduction in non-medical use to estimate the total number of avoided visits. To estimate the number of emergency department visits associated with abuse-deterrent extended-release oxycodone, avoided emergency department visits associated with extended-release morphine and transdermal fentanyl were subtracted from total avoided emergency department visits.

Avoided events and related cost savings were modeled for the entire US population using two different abuse-deterrent technologies, PC (low scenario) and agonist/antagonist (high scenario). Inputs for the model were primarily sourced from existing literature. Table 1 summarizes the inputs associated with the model.

Table 1. Key model inputs.

Input	Low scenario^a	High scenario^b	Source
Reduction in NMU rate	12.7% decline in NMU (physical barrier)	30.6% decline in NMU (agonist/antagonist)	White et al.^12
Diversion	15% of overall avoided events and costs	15% of overall avoided events and costs	Rice et al.^21, Yenikomshian et al.^10
Medical savings for oxycodone ER (Inflated to 2017 USD)	$515m	$1,236m	Rossiter et al.^9
Overall number of opioid related ED visits before ADO introduction	600,000	600,000	NIDA^16

^a Low scenario refers to the replacement of non-ADOs with ADOs with physical barrier technology.

^b High scenario refers to the replacement of non-ADOs with ADOs with agonist/antagonist technology.

Abbreviations. ADO, abuse-deterrent opioid; ER, extended-release; NIDA, National Institute on Drug Abuse; NMU, non-medical use.

Model outputs

Annual medical events and costs avoided were estimated for emergency department visits, outpatient visits, hospitalizations, substance abuse treatment programs and opioid abuse drug treatment, and injection-related diseases (HIV, Hepatitis-C). Excess prescription costs (combined opioid and non-opioid) were estimated. Indirect costs savings include disability and absentee days. Total cost savings include both direct and indirect costs.

Results

Based on a simulation of the US population over a 1-year period, extended-release abuse-deterrent opioids will lead to reductions in misuse and/or abuse-related events and costs. Table 2 presents annual medical events avoided and associated cost savings. Replacing all current extended-release morphine prescriptions with extended-release abuse-deterrent opioids resulted in 268,288–645,100 fewer medical events overall. In all scenarios, the largest proportion of avoided events was in outpatient settings (69%), followed by opioid abuse drug treatment facilities (18%), emergency department visits (7%), hospitalizations (4%), substance abuse-treatment facilities (1%), and injection related diseases (1%). Agonist/antagonist technology resulted in higher rates of avoided events and medical cost savings. Replacement of morphine with agonist/antagonist abuse-deterrent opioids led to direct and indirect cost savings of $394 M compared to $164 M with PC abuse-deterrent opioids.

Table 2. Estimated annual avoided medical events and costs.^a

		ER morphine		TD fentanyl
		Low scenario^b	High scenario^c	Low scenario	High scenario
Avoided events	ED visit	17,848	42,916	745	1,791
	Outpatient visit	183,803	441,954	7,673	18,449
	Hospitalizations	11,820	28,422	493	1,186
	Substance abuse treatment	3,073	7,390	128	308
	Opioid abuse drug treatment	48,699	117,097	2,033	4,888
	Injection-related diseases	3,045	7,321	63	153
Cost savings (millions of 2017 USD)	Direct: Medical	$120.63	$290.06	$4.88	$11.74
	Direct: Prescription drugs	$13.87	$33.36	$0.47	$1.12
	Indirect: Work loss	$29.35	$70.57	$1.23	$2.95
	Direct/indirect cost savings	$163.85	$393.99	$6.58	$15.81

^a Annual avoided medical events and costs were estimated over a 1-year time period for ER morphine and TD fentanyl for both low and high scenarios.

^b Low scenario refers to the replacement of non-ADOs with ADOs with physical barrier technology.

^c High scenario refers to the replacement of non-ADOs with ADOs with agonist/antagonist technology.

Abbreviations. ADO, abuse-deterrent opioid; ED, emergency department; ER, extended-release; TD, transdermal.

Replacing all current transdermal fentanyl prescriptions had a smaller impact on avoided events and medical cost savings. Replacing transdermal fentanyl prescriptions with extended-release abuse-deterrent opioids resulted in 11,135–26,775 fewer medical events and cost savings of $7 M to $16 M for low and high scenarios, respectively.

Table 3 shows the aggregate cost savings and annual avoided emergency department visits. In the high scenario, replacing extended-release morphine, extended-release oxycodone, and transdermal fentanyl with abuse-deterrent opioids results in over $1.5B in savings and 186,000 avoided emergency department visits. On average, replacing these non-abuse-deterrent opioids with PC abuse-deterrent opioids results in ∼40% of the cost savings associated with an agonist/antagonist replacement.

Table 3. Overall medical cost savings and avoided ED visits.

Opioid	Low scenario^a		High scenario^b
	Annual medical cost savings	Annual avoided ED visits	Annual medical cost savings	Annual avoided ED visits
ER morphine	$120m	17,848	$290m	42,916
TD fentanyl	$5m	745	$12m	1,791
ER oxycodone^c	$515m	59,407	$1,236m	141,293
Total	$640m	78,000	$1,538m	186,000

^a Low scenario refers to the replacement of non-ADOs with ADOs with physical barrier technology.

^b High scenario refers to the replacement of non-ADOs with ADOs with agonist/antagonist technology.

^c Medical cost savings for oxycodone were calculated from Rossiter et al.⁹. Avoided ED visits were calculated based on the total annual number of avoided ED visits from NIDA (2009). See also Table 1.

Abbreviations. ADO, abuse-deterrent opioid; ED, emergency department; ER, extended-release; NIDA, National Institute on Drug Abuse; TD, transdermal.

Discussion

The current study provides a deeper understanding of the potential real-world reduction in misuse and/or abuse-related medical events and costs that may be potentially achieved by replacing extended-release non-abuse-deterrent opioids with extended-release abuse-deterrent opioids. The cost savings observed with extended-release abuse-deterrent opioids in other studies persist in this current model, and leads to significant estimated reductions in medical events and direct/indirect cost savings. In addition to the cost savings and avoided events presented in Rossiter et al.⁹, Severtson et al.⁸ show that rates of extended-release opioid abuse exposures exhibit a significant decline following the introduction of an abuse-deterrent opioid⁸. A further review of data gathered after the introduction of reformulated OxyContin (abuse-deterrent oxycodone) by Coplan et al.²² demonstrated a 48% decrease in abuse of OxyContin in national poison center surveillance systems, a 32% decrease in abuse in a national drug treatment system, a 27% decrease in abuse in patients prescribed OxyContin, and a 65% decrease in reported overdose fatalities. The same decreases in abuse were not seen in other opioids without abuse-deterrent properties, further suggesting that the effects were due to the abuse-deterrent formulation and not broader opioid initiatives.

A recent draft evidence report released by the Institute for Clinical and Economic Review (ICER) on the effectiveness and value of abuse-deterrent opioids concluded that abuse-deterrent opioids on a hypothetical incident opioid population do not result in cost savings²³, despite confirming evidence that abuse-deterrent opioids result in markedly lower rates of real-world abuse⁸. ICER’s model predicts ∼2,300 fewer new cases of abuse and some $266 m in abuse-related savings in an incident population over 5 years. Apart from criticisms regarding the interpretation of primary literature, the ICER study most notably is modeled from a payer perspective and ignores any societal costs. The current study uses a different methodology than the one used in the ICER model by both considering indirect costs, including lost productivity and other societal costs, and also accounting for real-world non-medical use. This current model offers a more comprehensive assessment of the current US opioid environment and expands upon existing literature by incorporating the relationship between human abuse potential and non-medical use rates into this current model.

Since the first introduction of reformulated abuse-deterrent opioid oxycodone in 2010, adoption of abuse-deterrent opioids has been slow, despite strong FDA support for their development²⁴. As state governments become increasingly pressured to pass comprehensive drug reform, ∼30 state legislatures have introduced bills related to abuse-deterrent drugs and mandates involving insurance coverage of abuse-deterrent formulations²⁵. Yet, as of April 2017, bills have only been passed by five states²⁶. The findings from this current model suggest that expanding the availability and use of abuse-deterrent opioids may be another effective strategy to combat prescription opioid abuse and its consequences. However, the true effect of abuse-deterrent opioids can only be evaluated if all prescription opioids are made to carry abuse-deterrent properties. A working paper from the National Bureau of Economic Research found that reductions in the opioid supply can increase the use of abuse-prone substitute drugs such as heroin²⁷. Following the reformulation of abuse-deterrent opioid oxycodone in 2010, prescription opioid related abuse and overdoses declined while heroin overdoses more than tripled, suggesting a “balloon” effect where opioid abuse is shifted from one product to another as long as there are easily abused alternatives in the abuse ecology (e.g. non-abuse-deterrent opioids, heroin, illicit carfentanyl). If the availability of non-abuse-deterrent opioids remain widespread, patients prescribed abuse-deterrent opioids could easily obtain non-abuse-deterrent opioids from family members or through illegal channels, limiting the potential for abuse deterrence in the population.

Results for the current model are limited by the availability of certain data. Due to a lack of studies quantifying the relationship between abuse-deterrent opioids and increases in illicit drug use, the diversion percentage of 15% may differ from population values. One additional limitation stems from the incorporation of data from multiple human abuse potential studies from the White et al.¹² publication, as these data are not available by route of administration. The current study does not analyze prescription drug costs as part of the cost savings analysis, nor does the current study include additional indirect costs such as criminal justice costs associated with opioid misuse and/or abuse. Also, the cost calculations per avoided event in this current model are drawn from administrative insurance claims from a national population of privately insured patients, and may differ from actual costs incurred by other populations such as Medicaid or Medicare. Finally, we understand that the rate of opioid-related emergency department and inpatient visits have risen dramatically since 2009, the data year upon which the analysis for the current study was performed. Therefore, it is likely that this analysis under-estimates the reduction in these abuse-related events and their associated cost savings as a result of the replacement of non-abuse-deterrent opioids with abuse-deterrent opioids.

Conclusions

The current findings indicate that abuse-deterrent opioids may lead to a significant reduction in medical events utilization and work-loss costs. The replacement of all extended-release morphine prescriptions with an extended-release abuse-deterrent opioid led to significant reductions in medical events and direct/indirect cost savings, while the replacement of transdermal fentanyl prescriptions had a smaller impact. Extended-release abuse-deterrent opioids with agonist/antagonist technology were associated with a greater reduction in medical events and costs as compared to PC technology. The avoidance of these abuse related medical events yielded substantial annual medical cost savings of up to $1.5B for extended-release oxycodone, extended-release morphine, and transdermal fentanyl combined. Extended-release abuse-deterrent opioids serve an important role in effective pain management while lowering the risk of misuse and/or abuse. The current study contributes to the growing evidence base on the economic benefits of abuse-deterrent opioids, and suggests that a significant number of adverse health and healthcare consequences can be avoided by prescribing extended-release abuse-deterrent opioids in place of non-abuse-deterrent opioids, thereby reducing the societal burdens of opioid misuse and/or abuse.

Transparency

Declaration of funding

The current study was sponsored by Pfizer Inc.

Declaration of financial/other relationships

Research support was provided to Analysis Group, Inc. by Pfizer Inc. MY, AW, and MC are employees of Analysis Group, Inc., which received financial support from Pfizer in connection with the study and development of this manuscript. ZJ was an employee of Analysis Group, Inc. at the time of the study. CLR is an employee of Pfizer Inc. and owns stock in Pfizer Inc. MRM was an employee of Pfizer Inc. at the time of the study. No other conflicts of interest are directly relevant to the current study.

Previous presentations

Some results from the current study were presented in a poster at the American Pain Society 35th Annual Scientific Meeting, Austin, TX (May 2016) titled “A Model to Quantify Potential Medical Events Avoid and Cost Savings from Abuse Deterrent Opioids”.

Acknowledgements

None reported.